SaaS, PaaS, IaaS and Cloud Deployment Models

IaaS

Infrastructure as a Service supplies virtual compute, storage and networking while the customer manages the guest operating system, middleware.

PaaS

Platform as a Service supplies a managed application runtime and deployment platform while the customer manages application code, configuration and data.

SaaS

Software as a Service delivers a complete provider-operated application to users, usually through a browser or API.

public cloud

A public cloud exposes provider-owned shared infrastructure to many customers through accounts and isolated virtual resources.

private cloud

A private cloud supplies self-service pooled infrastructure for one organization, whether hosted on its premises or by a provider.

hybrid cloud

Hybrid cloud connects private and public environments with identity, networking.

packaged software and SaaS

Packaged software is installed and operated by the customer; SaaS is operated by a provider and accessed as a service.

service responsibility boundary

The division of operational, security and data-protection duties between a cloud provider and its customer.

cloud service selection

Choosing a cloud service according to workload behavior, control needs, security, availability, portability and total cost.

Cloud Service Models

Infrastructure as a Service supplies virtual machines, networks and storage. The provider operates facilities and virtualization while the customer manages the guest operating system, middleware, applications and data. Platform as a Service also manages the operating system and runtime, allowing the customer to deploy code. Software as a Service delivers a complete application and leaves the customer mainly responsible for users, configuration and entered data.

IaaS offers control but requires patching and runtime administration. PaaS reduces that work but limits supported behavior. SaaS provides rapid adoption with the least infrastructure control. Security remains shared: providers protect components they operate while customers control identities, permissions, data classification and safe configuration.

Deployment Models and Resilience

A public cloud uses provider infrastructure shared with logical isolation. A private cloud serves one organization. A hybrid cloud connects private systems with public services. Multi-cloud deliberately uses more than one provider. Selection depends on regulation, latency, existing systems, skills, cost and recovery needs.

Elasticity changes allocated resources as demand changes. Scalability is the ability to handle growth and does not guarantee automatic scaling. High availability uses redundant instances across independent zones. Disaster recovery restores service after larger loss and is defined through recovery time and recovery point objectives. Cost includes compute, storage, requests, licenses and network transfer.

Managed Services and Selection

Managed database, queue, cache and analytics services sit between raw infrastructure and complete SaaS applications. The provider operates backups, patching or clustering to the extent promised by the service while the customer still designs schemas, access and recovery objectives. A service-level agreement describes commitments and exclusions; an architecture must not assume behavior beyond the documented contract.

Workload placement begins with constraints. Data residency may restrict regions. Latency may keep processing near users or source systems. Existing licenses and skills affect cost. Dependency on a provider-specific API can be acceptable when its operational benefit is greater than the future migration cost. Containers standardize application packaging but do not make stateful services or network behavior identical across providers.

Control-plane operations configure resources while data-plane operations carry application traffic. A control-plane outage may prevent creating new capacity even while running instances continue. Region, availability-zone and service failure boundaries differ. Redundancy provides value only when copies do not share the same failure boundary and when routing can move traffic to a healthy copy.